Previous studies have related pair bonding in Microtus ochrogaster, the prairie vole, with plastic changes in several brain regions. However, the interactions between these socially-relevant regions have yet to be described. In this study, we used resting state magnetic resonance imaging to explore bonding behaviors and functional connectivity of brain regions previously associated with pair bonding. Thirty-two male and female prairie voles were scanned at baseline, 24h and 2 weeks after the onset of cohabitation. By using network based statistics, we identified that the functional connectivity of a cortico-striatal network predicted the onset of affiliative behavior, while another predicted the amount of social interaction during a partner preference test. Furthermore, a network with significant changes in time was revealed, also showing associations with the level of partner preference. Overall, our findings revealed the association between network-level functional connectivity changes and social bonding.

Thirty-two three-month-old sexually naïve female (N=16) and male (N=16) prairie voles (Microtus ochrogaster) were used in the study. The animals were housed in a temperature (23°C) and light (14:10 light-dark cycle) controlled room and provided with rabbit diet HF-5326 (LabDiet, St. Louis, MO, USA) oat, sunflower seeds, and water ad libitum. These voles were previously weaned at 21 days, housed in same-sex cages, and were descendants of voles generously donated by Dr. Larry J. Young from his colony at Emory University. The animal research protocols were approved by the bioethics committee of the Instituto de Neurobiología, UNAM.

Fourteen days before the experimental protocol, female voles were bilaterally ovariectomized. After recovery, silastic capsules (Dow Corning™ Silastic™ Laboratory Tubing; Thermo Fisher Scientific, Pittsburg, USA) containing estradiol benzoate (E2B; Sigma Aldrich, Missouri, USA) dissolved in corn oil (0.5 mg/mL of E2B) were implanted via s.c. to induce sexual receptivity four days before cohabitation protocol and remained implanted during the entire experimental protocol. Prairie voles underwent three MRI acquisition sessions: a baseline scan before cohabitation, a second scan after 24 hours of cohabitation, and a third scan after two weeks of cohabitation. The day after the baseline scanning session, female and male voles unrelated to each other were randomly assigned as couples and placed for cohabitation in a new home cage with fresh bedding to promote ad libitum mating and social interaction. Voles were housed in couples for the remainder of the experiment and were only separated for MRI scanning sessions and behavioral tests.

Animals were anesthetized to avoid stress and excessive movement during scanning sessions. Isoflurane at 3% concentration in an oxygen mixture was used for induction and positioning in the scanner bed, in which the head was immobilized with a bite bar and the coil head holder. Once voles were securely placed in the scanner bed, isoflurane anesthesia was adjusted at a 2% concentration and a single bolus of 0.05 mg/kg of dexmedetomidine (Dexdomitor; Zoetis, Mexico) was administered subcutaneously. Five minutes after the bolus injection, isoflurane anesthesia was lowered and maintained at 0.5%. MRI acquisition started when physiological readings were stabilized (~15 minutes after bolus injection). Body temperature was maintained with a circulating water heating pad within the scanner bed (Supplementary Figure 1), respiration rate was monitored with an MR-compatible pneumatic pillow sensor, and blood oxygen saturation was measured with an MR-compatible infrared pulse-oximeter (SA instruments Inc, Stony Brook NY). After the scanning sessions, animals were monitored until fully recovered and transferred back to their housing.

Prairie voles underwent three MRI acquisition sessions: a baseline scan before cohabitation, a second scan 24h after the onset of cohabitation, and a third scan 2 weeks after the onset of cohabitation. MRI acquisition was conducted with a Bruker Pharmascan 70/16US, 7 Tesla magnetic resonance scanner (Bruker, Ettlingen, Germany), using an MRI CryoProbe transmit/receive surface coil (Bruker, Ettlingen, Germany). Paravision-6 (Bruker, Ettlingen, Germany) was used to perform all imaging protocols. Before running the fMRI sequence, local field homogeneity was optimized within an ellipsoid covering the whole brain and skull using previously acquired field maps. rsfMRI was acquired using a spin-echo echo planar imaging (SE-EPI) sequence: repetition time (TR) = 2000 ms, echo time (TE) = 19 ms, flip angle (FA) = 90°, field of view (FOV) = 18 × 16 mm², matrix dimensions = 108 × 96, yielding an in-plane voxel dimensions of 0.167 × 0.167 mm², and slice thickness of 0.7 mm, total volumes acquired = 305 (10 minutes and 10 seconds). EPI bandwidth was 288461.5 Hz, the number of slices was 25, pi pulse (refocusing pulse) duration was 1.5455 ms with 2200.0 Hz bandwidth, and pi/2 (excitation pulse) was 1.9091 ms and 2200.0 Hz, respectively. After the rsfMRI sequence, an anatomical scan was obtained using a spin-echo rapid acquisition with refocused echoes (Turbo-RARE) sequence with the following parameters: TR = 1800 ms, TE = 38 ms, RARE factor = 16, number of averages (NA) = 2, FOV = 18 × 20 mm², matrix dimensions = 144 × 160, slice thickness = 0.125 mm, resulting in isometric voxels of size 0.125 × 0.125 × 0.125 mm³.

Imaging data preprocessing was performed with FMRIB’s Software Libraries (FSL, RRID:SCR_002823; Jenkinson et al., 2012) and Advanced Normalization Tools (ANTs; RRID:SCR_004757; Avants et al., 2011) for spatial registration. To avoid initial signal instability, the first 5 volumes of each functional series were discarded. Slice-timing correction and motion correction were applied using the first non-discarded volume as reference. The reference volume was also taken to determine the deformable transformation to the corresponding anatomical image. The resulting transformation was combined with a non-linear transformation to a prairie vole brain template obtained from previously published work (Ortiz et al., 2018). Functional images were later warped to the brain template and resampled to resolution of 0.16 × 0.16 × 0.7 mm3. To minimize physiological confounds, the first 5 eigenvectors (time-series) within the combined non-grey matter mask were obtained (Behzadi et al., 2007), since recent findings have shown that vascular, ventricle, and white matter signal regression enhances functional connectivity specificity from rsfMRI data (Grandjean et al., 2020). These eigenvectors and the 6 motion parameters (3 rotations, 3 displacements) were regressed out from each subject’s functional series. Datasets were band-pass filtered to retain frequencies between 0.01 and 0.1 Hz (Gorges et al., 2017). Finally, smoothing was applied with a box kernel with size of three voxels, using FSL (RRID:SCR_002823).

Due to technical problems, two subjects missed the baseline MRI acquisition (session 1), and two subjects missed the 2 week-cohabitation MRI acquisition (session 3). Additionally, two datasets of session 1 showed signal loss in the posterior cortex. The final sampling consisted of 90 datasets, with only 6 subjects missing one session.

README file for recommendations in data usage is also included.